Pion light cone wave function in the non-local NJL model

TPJU-4/2001

Pion light cone wave function in the non-local NJL model

Michal Praszalowicz and Andrzej Rostworowski

M.Smoluchowski Institute of Physics,

Jagellonian University,
Reymonta 4, 30-059 Krakow, Poland.
(February 1, 2008)

O
O

(N

<

ON

m
>
oo
oo

in
o

'

a:
i

X

We use the simple instanton motivated NJL-type model
to calculate the leading twist pion light cone wave function.
The model consists in employing the momentum dependent
quark mass in the quark loop entering the definition of the
wave function. The result is analytical up to a solution of a
certain algebraic equation. Various properties including the
kr dependence of the pion wave function are discussed. The
resulting kr integrated wave function is not asymptotic and
is in agreement with recent analysis of the CLEO data.

I. INTRODUCTION

Despite the fact that the exclusive hadron light cone
wave functions were theoretically introduced 20 years ago
0- 0, there is still relatively little experimental data
which could be confronted with theory. Recently the
analysis of Ref. || based on the latest CLEO measure-
ments H put some limits on the expansion coefficients
of the pion wave function in terms of the Gegenbauer
polynomials. This analysis indicates that the pion wave

function measured at
asymptotic

1.5 - 9.2 GeV 2 is neither

#»=6u(l-t0,

(1)

(with u being the fraction of the pion momentum carried
by the quark) nor of the form proposed by Chernyak
and Zhitnitsky in 1982 §,§. The first attempt to study
the kx dependence of the pion wave function has been
undertaken in Refs. ||.

In order to estimate the shape of the pion wave func-
tion at sub-asymptotic Q 2 , practically one theoretical ap-
proach has been used: QCD sum rules |M which gave
a variety of contradictory results ^,|n]- The extension
of the local sum rules to the case of the non-zero quark
virtuality proposed in Refs. |lj] has been recently suc-
cessfully applied to model the pion wave function [fl3|| ,
which turned out to be in agreement with the experi-
mental analysis of Ref. || •

There have been also lattice calculations of the mo-
ments of the pion wave function Jl4|,[l5| which, however,
gave unrealistically small results. Recent transverse lat-
tice results of Burkhardt et. al. [fL6| and Dalley [jl7J are in

contradiction. Dallcy's pion wave function is much more
flat than the one of Ref. g(J. It is, however, not very
different from our results for the constituent quark mass
M = 325 MeV.

Another approach has been recently put forward in
Refs. fl8|,|l9| where a simple instanton based model was
applied to calculate the u-dependence of the pion and
also photon wave function. The whole idea consisted in
calculating the quark loop which enters the expression
for the wave function, using the momentum dependent
constituent quark mass. The momentum dependence of
the quark mass appears naturally in the instanton model
of the QCD vacuum [^0|. It embodies the fact that the
propagator of a light quark gets modified by the vacuum
through which the quark propagates. The momentum de-
pendent quark mass acts as a natural cut-off function for
the fcr integration making this integration finite. Addi-
tionally, due to the non-local 7r-quark coupling, it ensures
vanishing of the pion wave function at the end points (see
Sect.IIIA). If the constituent quark mass is taken as a
constant and the sharp cut-off for the k T integration is
used, the pion wave function is simply a constant. For
more sophisticated cut-offs one can obtain more realistic
shapes, as in Refs. |2l| .

The instanton model of the QCD vacuum (for recent
review see e.g. p2|) allows to derive an effective non-
local Nambu-Jona-Lasinio (NJL) type quark-meson La-
grangian where mesons are not dynamical but composite
quark bilinears. By integrating out the quark fields one
can, however, generate the kinetic term for the mesons.
Quark-meson (pion in our case) interaction generated by
the instantons is non-local and this non-locality is param-
eterized by the momentum dependence of the constituent
quark mass. The constituent quark mass appears because
of the chiral symmetry breaking induced by the instan-
tons. However, the instanton vacuum is most probably
not able to explain the confinement.

In most applications of the effective quark-meson the-
ory described above an approximation of the constant
constituent quark mass was used. This approach has an
impressive success in describing baryon (which emerge as
solitons) properties (see [^2j and references therein) . Re-
cently an attempt to include momentum dependence in
calculating solitons in the non-local NJL model has been
reported in Refs. |23[ |.

The instanton model of the QCD vacuum is naturally

1

formulated in Euclidean space-time, whereas the light
cone wave function is an object defined in the Minkowski
space-time. One can in principle Wick rotate the un-
derlying integrals and perform the final integrals in the
Euclidean space-time as has been for example done in
Refs. @- [£(§ (for early attempts to calculate the pion
wave function in the instanton model see Ref. ^?J). Wc
will, however, perform all calculations directly in the
Minkowski space-time. The authors of Refs. |l^,[l9| used
a simple dipole type Ansatz for the momentum depen-
dence of the constituent quark mass in the Minkowski
space-time which made the calculations feasible. In Ref.
p8| an approximation was used where the momentum
dependence was taken into account only in the quark-
pion vertices, whereas in Ref. the calculation has
been done numerically. The results of both calculations
showed that the pion wave function at low momentum
scale was actually not very different from the asymptotic
one of Eq.(§).

In this paper we follow the approach of Refs. Jl8| , pj[ ,
however, we show how to calculate the pion wave function
analytically (up to the numerical solution of a certain al-
gebraic equation). We obtain compact, analytical results
both for u and k\ dependent wave functions. We use two
parameter set of Ansatze for the momentum dependent
quark mass which generalize the Ansatz of Refs. |l8| , |l9| .
In this way we can study the sensitivity of various prop-
erties of the pion wave function to the specific form of
the cut-off function. This is quite important, since - as
explained above - we are not able to use the momentum
dependent quark mass calculated in the instanton model,
but we model it in such a way that the calculations can
be performed in the Minkowski space-time.

A novelty of our approach is that we study the k T
dependence of the wave function. We calculate the first
moments in k T which are related to the mixed quark-
gluon condensates |^8| .

Our findings can be summarized as follows. The d 2 kx
integrated pion wave function is not asymptotic, it ex-
hibits a plateau for u ~ 1/2 whose height varies with the
cut-off function and the constituent quark mass at zero
momentum. Comparing the coefficients of the expansion
in terms of the Gegenbauer polynomials, a2k, with the
analysis of the CLEO data || we find that our wave func-
tion fits within 95% confidence level in the a-i — 04 two di-
mensional parameter space, whereas the asymptotic and
the Chernyak-Zhitnitsky wave functions are ruled out.
One should note, however, that in the most cases the
coefficient ae is in our case not negligible, while it was
explicitly put to zero in Ref. [5JL In this respect we also
differ from the model of Ref. [ 13j .

The k T dependence of the u integrated wave function,
for k T below 1 GeV 2 , is insensitive to the considered
forms of the cut-off function. In this region, the wave
function can be well fit by C(k T + A 2 )~ Q with a ~ 8 -j-
9. For k T < 0.2 GeV 2 the exponent fit can be used
as well. The asymptotic power like behavior depends
strongly upon the form of the cut-off function. However,

the difference can be seen only for k\ above few GeV 2 ,
whereas the region below 1 GeV 2 contributes 98% to the
norm of the wave function. It indicates that the Fermilab
experiment E791 || where the typical quark (jet) kx is
above 1 GeV, is in fact sensitive to the tiny portion of the
pion wave function. The ratio R —< k% > / < k T > 2 is
of the order of 3 indicating rather narrow k T distribution
in contrast to Ref. |^8| where R ~ 5 — 7.

We find that, given simplicity of the model, both con-
ceptual and technical, our results are reasonable and,
where the comparison with the data is possible, sur-
prisingly accurate. The method used in this paper can
be straightforwardly generalized to other wave functions
like, for examp le, ph oton wave function (which we briefly
discuss in Sect. Ill A ) or 2tt distribution amplitude.

The paper is organized as follows: in Section [n| we show
how to calculate the quark loop with the momentum de-
pendent mass analytically. The main problem here is,
how to handle the complex poles of the quark propagator.
In Section III we discuss various properties of the pion

wave function, like end point behavior and kx asymp-
totics. We also present numerical results and compare
with other models. Finally in Section IV we present our
conclusions.

II. PION WAVE FUNCTION WITH MOMENTUM
DEPENDENT QUARK MASS

Let us consider two quarks moving along the light cone
direction n = (1, 0, 0, 1) parallel to the total momentum
P and separated by the light cone distance z — 2r along
the direction n = (1,0,0,-1). The twist 2 light cone
pion wave function is then defined as

1

eh

-ir(2u-l)(nP)

— OO

I $(nT)fi~f 5 ip (-nr) \tt + (P))

(2)

where variable u, as will be shortly seen, has a meaning
of a fraction of a quark momentum k + with respect to
P + . Here = 93 MeV. In this kinematical frame any
four vector v can be decomposed as:

(3)

with v + = n ■ v, v = n ■ v and the scalar product of
two four vectors reads:

1 + -

1 - h

—v w
2

(4)

In Eq.(^|) the path ordered exponential of the gluon field,
required by the gauge invariance, has been omitted since
we shall be working in the effective quark model where
the gluon fields have been integrated out.

2

The instanton model of the QCD vacuum predicts that
quarks interact non-locally with an external meson field

u MM

St = M

d 4 kdH

ip(k)F(k)U 1B (k-l)F(l)ip(l) (5)

and C/ 75 (x) can be expanded in terms of the pion fields

f

E7*« = 1 + irl b r A TT A
F w

2Fl

IT 7T

(0)

M is a constituent quark mass of the order of 350 MeV
and F(k) is a momentum dependent function such that
F(0) = 1 and F(k 2 -> oo) -> 0.

I k I [GeV]

FIG. 1. F(k) for Euclidean momentum k 2 < 0, for n = 1
(long dashed), 3 (dashed) and for the instanton model (solid).

Function F(k) has been calculated in the instanton
model of the QCD vacuum

F mst .(k) = 2z \I (z)Ki{z) - h(z)K Q (z)} - 2I l {z)K l {z)

(7)

where z = kp/2. Here 1/p ~ 600 Mev is a typical inverse
instanton size. One has to remember that this function
is defined in Euclidean space as a Fourier transform of
the instanton zero mode. Here, following Refs. [ p~8|Jl9|
we wish to calculate <f>^ directly in the Minkowski space.
To this end we shall choose a simple pole formula

M{k) = MF 2 (k)

with

F(k)

-A 2

k 2 - A 2

(8)

which reproduces reasonably well (^) for k 2 < 0. This is
shown in Fig.[j] where we have chosen A in such a way that
(||) crosses with (^) for k — 1.5 GeV. Unfortunately large
k asymptotics Fi nst (k) ~ k~ 3 cannot be reproduced by
Eq.(^|) for integer n. Choosing half- integer n, however,
would introduce cuts in the complex k plane, making the
numerical evaluation of the pion wave function unneces-
sary tedious. The cutoff parameter A has to be chosen

in such a way that the pion wave function satisfies the
normalization condition

du0 T (u) = 1.

(9)

Matrix element (0) reduces to a simple fcrmion loop

, , . iN c f d 4 k ( k+

<t>ir(U = o / jO\U

i F 2 P+ j (27r)4 ^ P+

Tr

, ijM(k) iJM(k-P)

'fc-M(k) 10 {t-f 3 )- M{k)

(10)

Here it is explicitly seen that uP + = k + . In the light-
cone representation (||) d A k — ^dk + dk~ d 2 ^ and after
performing the trace we get

2N C M 2 P+ f d 2 k T

F 2

dk.

(2n) 2 J (2tt) 2

F(k)F(k~P)

(l-u)F 2 (k) + uF 2 (k- P)

[k 2 - M 2 F i (k) + ie][(k - P) - M 2 F i (k - P) + ie]

(11)

In order to calculate the integral over <ifc_ exactly i.e.
with full k dependence in the denominators we have to
find zeros of the two propagators in Eq. (|ll|) . To this end
it is convenient to introduce the scaled variables

P +

P+ =

A ' A '

ri=p+K-, t=(K T ) 2 , r 2

In these variables

N M 2 r+oo

T'

M 2
A 2 "'

(12)

4>n(u) = -i

(2tt) 3 F 2 J q

dt

drj

(i - u) cr ci" + u Ci 3ra q

G(Ci) G(Ca)

(13)

where

Q l = u I Tj h ie sign(u /

t+1

— ie sign(l — u

and

4n+l

G(z) = z in+l + z 4n -r 2 = l\ (z-Zi).

i=i

with Zi being (4n + 1) roots of the equation
G(z) = 0.

(14)

(15)

(16)

3

We aim at evaluating the drj integral by contour integra-
tion. The 2(4n+ 1) simple poles of the integrand (|13| ) lie
at

(i)

(2)

t+l + Zj

u

t+l + Zj

1 - u

it sign(ttj,

+ iesign(l — u). (17)

Let's assume for definiteness that < u < 1. For r 2 =
(A 2 = oo) the roots of Eq. ( [l6| ) are z\ = ... = z^ n = 0,
Z4„+i = —1 so the poles r/ 1 ) lie below and the poles ?/ 2 )
lie above the Ken axis. When r 2 increases (A 2 decreases
from oo to its physical value) the roots z% drift in the
complex z-plane as shown in Fig. |^. Therefore some
of the poles n^ can move above and some of the poles

(2)

rfl can move below the Rery axis. The crucial point
is that if we kept the integration path going along the
Re?y axis we would get an unphysical result: ff (u) would
not vanish outside the interval Q < u < 1. We assume
that physically correct prescription for performing the dn
integral is to deform the integration path as shown in Fig.

Rez

FIG. 2. Flow of the zeros of G(z)with increasing r for
n = 2. For r — (A = oo) there are 4n degenerate solu-
tions zi, Z4n and one non degenerate solution Z4n+i = — 1.
With increasing r (A decreasing from oo to the physical value)
Zttn+i solution drifts rightwards toward 0, and An zero solu-
tions split into 2 real ones, Z4„ drifting towards Z4 n +i and Z\
towards +oo, and 4n — 2 complex (pairwise conjugated) solu-
tions Z2i, Z2»+i, * = X, In — 1 which flow into the upper and
lower parts of the complex z plane. When r reaches the value
of the local maximum r max of G(z) for — 1 < z < the two
real zeros Z4 n and Z4 n +i meet and "scatter" drifting further
into the upper (z4 n ) and lower (z4 n +i) half of the complex z
plane.

Closing the contour in the upper complex r\ half-plane

(2)

we should enclose all the poles n\ and none of the poles

n^. This prescription ensures that (p„(u) is real and
vanishes outside the interval < u < 1. Now, the dr\
integral yields

n m 2 r +co 4n+1

MU) = (2^Ff (_1 + UT I dt ^ Z * fi (18)

U(t + 1 + UZif n + (1 - u) 2n+1 Z 2n {t + 1 + UZi) n

nti'^+i+^+ii-^] '

where factors /j are defined as

4n+l

Zk %i

in+1 ,

fi= n —

(19)

k=l
k^i

Re?7

FIG. 3. Schematic flow of the r\ poles with increasing r for
n — 2 together with the integration contour in the complex r\
plane. For r — the poles are shown as black dots and the
integration contour is represented by the dotted line.

With the help of the following identity
u(t + 1 + uzi) 3n + (1 - u) 2n+1 z 2n (t + 1 + uz,) n

i 4n+1 7/ 7 2n -i-n n\7 2n

E «gA : " *■ ^ — \,- (M)

(-1 + w)''

t + 1 + UZj + (1 - 1t),Zfc

fe=i

the result for the /c|>-dependent pion wave function
&(u, fc^) can be cast in the following compact form:

1 N C M 2
~ A 2 (2tt) 2 F 2

+ l zf ^\ + zf n z%(l - u)

E ■

i,fe=l

S + 1 + Ziit + z fc (l - u)

(21)

Factors /, obey the following properties which are crucial
for the finiteness of the d(kj<) integration and for the
end point behavior of the pion (and also photon) wave
function

4n+l

E z Th

i=l

for m < An,

1 for m = An.

(22)

4

It is interesting to note that properties (22) hold for any
set of N = 4n+ 1 numbers, irrespectively of the fact that
they are solutions of certain equation of degree N.

Due to Eq.(22) ^^(u, k^,) vanishes for large k\ and the
d(fc^) integration is finite so that

M = 325 GeV

&r(u) = / d(fer0$7r(u,fcr')

/i/ fc (^«+«r«i?(i-«))

x ln(l + z^it + z fe (l - u)) . (23)
Eqs.( pT J23 ) are our final results for the pion light cone

wave function for the momentum dependent quark mass
(|J). These analytical formulae depend on the numerical
solutions Zi of Eq. ([l6]) . Cutoff parameter A can be found
by imposing the normalization condition (0).

III. PROPERTIES OF THE PION WAVE
FUNCTION

A. Analytical properties

Function (j) v {u) = (f> v (l — u) which is a trivial property
of the symmetry of exchanging two sums over i and over
k in Eq.(|23]). At the end points ^(u) vanishes. Indeed

N AT 2

M ° ] = ~ (2^pi^ ? fiZ * n E ln ( J + Zfe ) =

(24)

as a consequence of Eq. ( p2[ ) . In fact for small it (u) ~
it". The reason for this kind of behavior is the presence
of the factor y/M(k)M(k - P) in Eq.©. Indeed, one
can calculate in the same way the transverse photon wave
function <j>^ (u) which, since the 7-quark coupling is local,

does not contain the dumping factor yj M(k)M(k — P).
The result reads

, N C M x ^

^ (u) = _ (2^7 T ^ /

2n _4n

'T z,fc

x ln (1 + + Zfe(l — it))

(25)

and the wave function <^r(0) = </>~(l) 7^ 0, due to the
second property (p2|). So despite of the nonperturba-
tive effects embodied in the momentum dependent quark
mass (||) the point like photon-quark coupling makes the
photon wave function non vanishing at the end points.

The advantage of the present calculation is that we
can calculate not only the it dependence, but also the
k\ dependence of the pion wave function. Integrating
3>7r(it, fcy) over u we get

M = 350 GeV

3

-e-

/

"\

i *s

f 1
//

hi

\ \

\\

1:1

- ///

\\\ .

»,

I'll
t!

\
i

0.2 0.4 0.6 o.e

u

M = 375 GeV

M = 400 GeV

<}>„(]&)

du $?tt(u, kj-)

(26) FIG. 4. Pion wave function for M = 325, 350, 375 and

400 MeV, for n = 1 (solid), 2 (long dashed) and 5 (dashed)
together with the asymptotic wave function (dot-dashed).

5

By expanding cj>(k^,),

Eq.(26), for large k% we find

M = 350 GeV

4>„(k T )

4n+l

(27)

This kind of behavior differs from the commonly assumed
exponential form[j] of ^(kj,). Let us, however, remark
that the asymptotic behavior (^7|) which switches on for
k\ > A 2 concerns only a negligible tail of the whole wave
function. Moreover, the power in Eq.(27) is large enough,
even for n = 1, to make the predictions of the first k\
moments reliable. We shall come back to this point in
the next Section.

B. Numerical results

In Fig.| we plot (f> n (u) for M = 325, 350, 375, 400 MeV
and n = 1,2 and 5. In fact for n > 5 4>tt(u) does not
change much anymore. It is clearly seen from these
figures that in all cases the NJL model with the non-
local regulator (||) gives 0^ (u) which is different from the
asymptotic one. This difference is more pronounced for
smaller masses M and larger n's. For M — 325 and
350 Mev and n — 5 a, shallow minimum can be seen at
u = 0.5.

It is instructive to compare our result with the one
of Refs. |l8|,|l9| where the momentum dependence of the
quark mass has been taken into account only in the nu-
merators of Eq.(pd|). This is done in Fig.|| for M = 350
MeV and n = 1. It is clearly seen that the momentum
dependence of the denominators flattens the pion wave
function making it " less asymptotic" .

TABLE I. Gegenbauer coefficients

M

n

A

2

04

06

MeV

MeV

1

1249

0.1363

0.0215

0.0037

0.0007

325

2

1862

0.0935

-0.0183

-0.0155

-0.0077

5

3033

0.0606

-0.0493

-0.0217

-0.0060

1

1156

0.1144

0.0151

0.0015

0.0001

350

2

1727

0.0659

-0.0271

-0.0166

-0.0064

5

2819

0.0284

-0.0600

-0.0202

-0.0020

1

1081

0.0931

0.0113

-0.0003

-0.0002

375

2

1621

0.0387

-0.0325

-0.0166

-0.0047

5

2649

-0.0035

-0.0668

-0.0173

0.0025

1

1020

0.0723

0.0099

-0.0020

-0.0001

400

2

1534

0.0118

-0.0347

-0.0164

-0.0027

5

2512

-0.0351

-0.0695

-0.0138

0.0072

FIG. 5. Pion wave function for M = 350 MeV, for n = 1
(solid), compared with the approximate result of Ref.[17]
(dashed) and the asymptotic wave function (dot-dashed).

The comparison of different wave functions and espe-
cially their content of the non-asymptotic terms is read-
ily done by expanding <p„ (u) in terms of the Gegenbauer
polynomials:

^r(u) = 4>l s {u) [l + a 2 Cl ,2 (2u - 1) + ai Cl /2 {2u - 1)
+a 6 C* 6 3/2 (2 U .-l) + ...l (28)

where the coefficients a 2 k depend on the scale /x 2 . In
model calculations it is, however, hard to define precisely
what value should be taken for /i. Clearly /i is of the
order of the cutoff A. In Table | we show the cutoff
A and a % ,... ,a s for M = 325,350,375,400 MeV and
n = 1, 2 and 5. It is clearly seen that in most cases is
still comparable with a^. Only as is always substantially
smaller than a 2 , 04 or a^. Interestingly for M = 325 MeV
and n = 1 we get a 2 very similar to the transverse lattice
result of Dalley JT§ who gets a l 2 atL = 0.133.

As already stated in the previous sections the non-local
NJL model gives the pion wave function as a function of
two variables u and k\. In Fig.^Jwe plot ^(k^) defined
in Eq.(|6|) for M = 350 Mev and n = 1,2 and 5. These
functions are indistinguishable within the accuracy of the
plot. </),r(&y) can be well approximated by the function
C{k\ + A 2 )-", with a = 9, A = 1 GeV 2 and C = 8
GeV 16 . The n dependence is visible only for k\ above a
few GeV 2 and is not taken into account in this fit.

In order to illustrate the (un) importance of the asymp-
totic power like behavior, let us define

Af(\) = / rf(fc 2 )^(fc 2 ).

(29)

1 The exponential fit, however, is quite satisfactory for fc^ <
0.2 GeV 2 .

Obviously Af(oo) = 1. However, JV(X) saturates rela-
tively fast

Af(l GeV) - 0.98

(30)

G

for all masses and powers of n. In fact the higher n
the larger value of N(l GeV). This observation indi-
cates that the bulk of the wave function comes from the
region where there is little, if any, dependence on the cut-
off parameter n. Outside this region the /^-dependent
wave function starts to depend strongly on the form of
the cutoff function (||). Unfortunately the fcr range in
the recently measured diffractive dijet production in the
pion nucleus scattering ||] is precisely in this asymptotic
regime. Therefore we have to conclude that the informa-
tion provided by this piece of data cannot be used to test
our model, unless the range is not shifted towards the
MeV rather than GeV region.

M = 350 GeV

kl[Ge\]

FIG. 6. 4>n(k^) for M = 350 and n = 1,2 and 5. These
curves are indistinguishable within the scale of the plot.

It was argued in Ref. [£8|] that the QCD sum rules predict
R ~ 4 — 5. In our case R is almost two times smaller,
however the absolute value of < k\ > is slightly larger
than the one of Ref. . Let us remind that the kj, mo-
ments are related to the mixed quark-gluon condensates
which arc poorly known in QCD.

M = 350 GeV

0.2

0.4 0.6 0.8 1

FIG. 8. Ratio < k\ > (it)/&r(u) for M = 350 MeV, and
n = 1 (solid), 2 (long dashed) and 5 (dashed). If there was
factorization, this ratio should not depend upon u.

We have checked the hypothesis that Q(u, k^) factor-
izes:

^^(u, k\) = 7r (u)0 Tr (fc|).

(31)

If this was the case we should get < k\ > (m)/(/) w (m) —
const =< k\ >. The actual plot of < k\ > (w)/</> w (u)
dependence is presented in FigM

M = 350 GeV

0.2

0.8 1

FIG
and n

7. < fcy > in pion wave function for M = 350 MeV,
= 1 (solid), 2 (long dashed) and 5 (dashed).

In order to see how the transverse momentum is dis-
tributed among quarks with different longitudinal mo-
mentum it is instructive to plot < k\ > as a function of
u. This is depicted in Fig.fi]. One can see that < k\ >
saturates relatively fast and remains almost constant for
0.2 < u < 0.8. The saturation value decreases with an in-
creasing M. In Table average < k\. > and < k^ > are
displayed together with the ratio R —< k^, > 2 / < k\ >.

C. Comparison with the data

Expansion of the pion wave function in terms of the
Gegenbauer polynomials has been defined in Eq.(|2"§|)
and the values of the expansion coefficients for (j) v {u) of
Eq. ( p3| ) are displayed in Table B. Schmedding and

TABLE II. The values of < kl >.

< k% > and ^r-

<*4> 2

M n
MeV

< fcf >
GeV 2

< fci >
GeV 4

< k% >/< k

2 ^2
T >

1

325 2
5

350

1

375 2

5

400

0.451L
0.433) 2
0.424) 2

0.203
0.188
0.180

0.437f
0.420) 2

0.191
0.176
0.411)^ = 0.169

0.426) 2
0.409) 2
0.402) 2

0.182
0.167
0.162

0.418) 2
0.402) 2
0.394) 2

0.175
0.162
0.155

0.620) 4
0.564) 4
0.542) 4

0.148
0.101
0.086

3.58
2.90
2.67

0.593)^"
0.541) 4
0.520) 4

0.124
0.086
0.073

3.39
2.76
2.55

0.572)^
0.522) 4
0.503) 4

0.107
0.074
0.064

3.24
2.65
2.45

0.555) 4
0.508) 4
0.489) 4

0.095
0.067
0.057

3.11
2.55
2.37

7

Yakovlev || performed recently a theoretical analysis
of the latest CLEO ||] data on pion electroproduction
from a quasi real photon. They constructed a 95% and
68% confidence level contour plots in the a 2 — cii parame-
ter space for an average virtuality /i = 2.4 GeV. In Fig.^
we present their contours (Fig. 6 in together with our
values of 02 and 04 for various fits displayed in Table Q.
We have not performed the QCD evolution of these co-
efficients since it is not clear which scale should be taken
as a starting point of the evolution. It is naturally to
assume that [jlq ~ A. In that case, however, the evolu-
tion would be very weak, since - depending on the cut-off
function - A changes between 1 and 3 GeV. As the result
of the evolution the points in the ai — 0,4 plane would
slightly move towards the asymptotic wave function.

Let us finally note, that the analysis of Ref. || has
been performed under the assumption that ag = 0. As
can be seen from Table | for n > 1 ae is still comparable
with 04.

0.1

0.0

-0.1

-0.2

-0.3

.1/ =

o 10 o 10

O N iO !N
TP CO CO CO

asymptotic ,. "~ \ ,-»^n = 1
.■■ V" *' n — 5

Ch.-Zh.

0.0

0.1

0.2

0.3

0.1

FIG. 9. The parameter space (02,04) of Ref.[5]. Black
dots represent different model predictions, solid contour cor-
responds to 68% confidence level, whereas the dashed one to
95%.

IV. SUMMARY AND CONCLUSIONS

In this paper we have calculated both u and k\ de-
pendence of the light cone twist 2 pion wave function in
the non-local NJL model. The nonperturbative nature of
this approach is embodied in the momentum dependence
of the constituent quark mass. In principle this depen-
dence is known from the instanton model of the QCD
vacuum. However, in order to be able to carry out the
calculations in the Minkowski space-time we have used
a class of simple Ansatze (@), which for Euclidean mo-
menta relatively well reproduce Mi ns t.(k). In this way
we have constructed a transparent and tractable model
which allowed to calculate the pion wave function (u)
analytically. We have shown how for the finite cut-off
parameter A the Feynman integration contour has to be

modified in order that (f> n (u) is real and vanishes outside
< u < 1.

Our phenomenological results are encouraging. For
the constituent masses at zero momentum between M =
325 - 350 MeV and n = 2 (see Eq.(|)) our results nicely
fit into the 95% confidence region of Ref. J 5|. In compar-
ison with the as ympto tic wave function (n ) and with the
results of Refs. [^8||l9|] where the momentum dependence
of the constituent mass in the denominator (c.f. Eq.(ll))
was neglected, our wave function is broader with, how-
ever, less pronounced end point regions. Similar behav-
ior was found in Ref. |l3| where not only non-locality
(within the sum rules approach) but also radiative cor-
rections have been taken into account. The fact that the
true pion distribution amplitude may be broader than
the asymptotic one has been already pointed out in Ref.

Although at first sight we have found rather strong
sensitivity to the power n in the fc|> distribution, it
turned out that the power like asymptotic tail <p^(k^) ~
(l/fcy) 4n+1 switches on at rather high momenta of the
order of a few GeV. For small and moderate k\ there
is no dependence of ^(k^) on n. We have shown that
more than 98% of the norm of (f>„(kj,) is concentrated in
the region of k\ < 1 GeV 2 . Therefore recent measure-
ments of the diffractive dissociation of a pion into 2 jets
with transverse momentum above 1 GeV Q test the tiny
portion of the wave function. Most probably, since the
higher kj- moments are divergent for the power like tail,
this region is not reliably described by our model.

Another way to extract a light cone wave function
would be to calculate the entire physical process within
the effective model, rather than only the wave function
itself as given by Eq.(B). This approach has been advo-
cated recently in Ref. [p6| and for the process 77 — > tt
can be in principle carried out also within our approach.
It would be interesting to see what is the difference be-
tween the wave function calculated directly from Eq.(j^)
and according to the prescription of Ref. M .

As a by-product we have also calculated the transverse
photon wave function 4>^{u) which is almost insensitive
to the parameter n of the constituent quark mass. Fur-
ther application of our method to strange mesons and, for
example, 2ir distribution amplitudes is straightforward
and will be discussed elsewhere. It is also possible by
calculating pion wave functions with other Dirac struc-
tures in (0) to disentangle different condensates related
to the fcy moments |2S} |.

This work was partially supported by the Polish
KBN Grant PB 2 P03B 019 17. M.P. is grateful to
W.Broniowski, K.Goeke, P.V.Pobylitsa, M.V.Polyakov
and N.G.Stefanis for discussions and interesting sugges-
tions. The authors kindly acknowledge comments and
remarks of A.Bakulev, A.Dorokhov and T.Hcinzl.

8

V.L. Chernyak and A.R. Zhitnitsky, JETP Lett. 26
(1977) 359.

S. Brodsky and G.P. Lepage, Phys. Lett. B87 (1979) 594;
Phys. Rev D22 (1980) 2157.

G. Farrar and D. Jackson, Phys. Rev. Lett. 43 (1979)
246.

A.V. Efremov and A.V. Radyushkin, Theor. Mat. Phys.
42 (1980) 97; Phys. Lett. B94 (1980) 245.
A. Sch meddmg and O. Y akovlev, Phys. Rev D62 (2000)
116002, frep-ph/9905392 .

J. Gronbe rg (CLEO Coll aboration), Phys. Rev D57

(1998) 33, |hep-ex/9707031
V. L. Chernyak and A. R. Zhitnitsky, Nucl. Phys. B201
(1982) 492; ibid B214 (1983) 547(E).
V.L. Chernyak and A.R. Zhitnitsky, Phys. Rep. 112
(1984) 173.

E.M. Aitala et.al. (Fer milab E791 Colla boration), |hcp-
ex/0010043] D. Ashery, ^iep-ex/9910002[ 4.
M.A. Shifman, A.I. Vainshtein and V.L Zakharov, Nucl.
Phys. B147 (1979) 385.

V.M. Braun and I.E. Filyanov, Z. Phys. C44 (1989) 157.
S.V. Mikhailov and A.V. Radyushkin, JETP Lett. 43
(1986) 712; Sov. J. Nucl. Phys. 49 (1989) 494; Phys.
Rev. D45 (1992) 1754.

A.P. Bakulev, S.V. Mikhailov and N.G. Stefanis,
Ruhr-Unive rsity preprint 2001, RUB-TPII-03/01

ph/0103119

nep-

D. Daniel, R. Gupta and D.G. Richards, Phys. Rev. D43
(1991) 3715.

L. Del Debbio et. al., Nucl. Phys. Proc. Suppl. 83 (2000)

235, |hep-lat/9909147 .
M. Burkardt and H

(1999) 054504, hep-ph/9805495

El-Khozondar
M

hep-ph/0101338
S. Dalley, frep-ph/0010043

Phys. Rev.
Burkardt and S.

D60

Seal,

V.Yu. Petrov and P.V. Pobylitsa, |hcp-ph/9712203 .
V.Yu. Petrov, M.V. Polyakov, R. Ruskov, C. Weiss
and K. Goeke, Phys. Rev. D59 (1999) 114018, Ihep

ph/980722£

D.I. Diakonov and V.Yu. Petrov, Nucl. Phys. B245
(1984) 259; B272 (1986) 457.
T. Heinzl, L iqh-Cone Quant ization:
Applications, |hep-th/0008096|; Nucl.

Foundations ana
Phys. Proc. Suppl.

90 (2000) 83, |hep-ph/00083l4
D.I. Diakonov and V.Yu. Petrov, |hep-ph/000900e .
B. Golli, W. B roniowski and G . Ripka, Phys. Lett. B
437 (1998) 24, |hep-ph/980726l| ; G. Ripka and B. Golli,

J. Nucl. Phys. 51

hep-ph/9910479

S.V. Esaibegyan, S.N. Tamarian, Sov.
(1990) 310 [Yad. Fiz. 51 (1990) 485].
A.E. Dorokhov and L. Tomio, Phys.Rev. D62 (2000)
014016.

I.V.Anikin, A.E. Dorokhov and L. Tomio, Phys. Lett.

B475 (2000) 361, |hep-ph/990936£ .

V.B. Rosenhaus, M.G. Ryskin Sov. J. Nucl. Phys. 44
(1986) 829 [Yad. Fiz. 44 (1986) 1276].
A.R. Zhitni tsky, Phys. Lett. B329 (1994) 493, |hep

ph/940127£

N. G. Stefanis, W. Schroers and H.-Ch. Kim, Eur. Phys.

J. C18 (2000) 137, hep-ph/000521£